def test_law_node_finder(self):
finder = LawNodeFinder(self.build_sample_law_tree())
nodes = finder.find(Query.from_text('第一条第二項'))
self.assertEqual(1, len(nodes))
node = nodes[0]
self.assertTrue(isinstance(node, Paragraph))
self.assertEqual('第一条第二項', node.sentence)
def test_apply_replace(self):
node = Paragraph(title='第一項', sentence='私はネコです')
node_finder = LawNodeFinder([node])
action = parse_action_text('第一項中「ネコ」を「イヌ」に改める')
apply_replace(action, node_finder)
self.assertEqual('私はイヌです', node.sentence)
def test_apply_delete(self):
node = Paragraph(title='第一項', sentence='私はネコです')
node_finder = LawNodeFinder([node])
action = parse_action_text('第一項中「私は」及び「です」を削る')
apply_delete(action, node_finder)
self.assertEqual('ネコ', node.sentence)
def test_apply_add_word_fail(self):
node = Paragraph(title='第一項', sentence='私はネコです')
node_finder = LawNodeFinder([node])
action = parse_action_text('第一項中「サル」の下に「ザメ」を加える')
with self.assertRaises(TextNotFoundError):
apply_add_word(action, node_finder)
def test_apply_add_word(self):
node = Paragraph(title='第一項', sentence='私はネコです')
node_finder = LawNodeFinder([node])
action = parse_action_text('第一項中「ネコ」の下に「ザメ」を加える')
apply_add_word(action, node_finder)
self.assertEqual('私はネコザメです', node.sentence)
def test_apply_replace_multiple_fail(self):
node = Paragraph(title='第一項', sentence='私はネコネコです')
node_finder = LawNodeFinder([node])
action = parse_action_text('第一項中「ネコ」を「イヌ」に改める')
with self.assertRaises(MultipleTextFoundError):
apply_replace(action, node_finder)
def test_apply_delete_fail(self):
node = Paragraph(title='第一項', sentence='私はネコです')
node_finder = LawNodeFinder([node])
action = parse_action_text('第一項中「私は」及び「でした」を削る')
with self.assertRaises(TextNotFoundError):
apply_delete(action, node_finder)
self.assertEqual('私はネコです', node.sentence) # not changed
def main(law_fp, gian_fp, out_fp, stat_fp, applied_fp, failed_fp, skipped_fp):
LOGGER.info(f'Start to parse {law_fp}')
try:
meta = extract_law_meta(law_fp)
nodes = parse_xml_fp(law_fp)
except Exception as e:
msg = f'failed to parse {law_fp}: {e}'
LOGGER.error(msg)
sys.exit(1)
node_finder = LawNodeFinder(nodes)
if gian_fp:
LOGGER.info(f'Start to apply {gian_fp}')
stats_factory = StatsFactory(['file', 'process', 'success'])
applied_actions, failed_actions, skipped_actions = apply_gian(gian_fp, node_finder)
process_count = len(applied_actions) + len(failed_actions)
success_count = len(applied_actions)
stats_factory.add({'file': gian_fp, 'process': process_count, 'success': success_count})
LOGGER.info('Applied {} / {} actions'.format(success_count, process_count))
if stat_fp:
stats_factory.commit(stat_fp)
LOGGER.info(f'Appended stats to {stat_fp}')
if applied_fp:
with open(applied_fp, 'w') as f:
for action in applied_actions:
f.write(json.dumps(action.to_dict(), ensure_ascii=False) + '\n')
LOGGER.info(f'Saved applied actions to {applied_fp}')
if failed_fp:
with open(failed_fp, 'w') as f:
for action in failed_actions:
f.write(json.dumps(action.to_dict(), ensure_ascii=False) + '\n')
LOGGER.info(f'Saved failed actions to {failed_fp}')
if skipped_fp:
with open(skipped_fp, 'w') as f:
for action in skipped_actions:
f.write(json.dumps(action.to_dict(), ensure_ascii=False) + '\n')
LOGGER.info(f'Saved failed actions to {skipped_fp}')
save_law_tree(meta['LawTitle'], nodes, out_fp)
LOGGER.info(f'Saved result to {out_fp}')
def test_law_node_finder_fail(self):
finder = LawNodeFinder(self.build_sample_law_tree())
self.assertEqual(0, len(finder.find(Query.from_text('第一条第一号イ'))))
self.assertEqual(0, len(finder.find(Query.from_text('第三条'))))